As the size of functional elements in integrated circuits decreases, complexity and interconnectivity increases. To accommodate the growing demand of interconnections in modern integrated circuits, on-chip interconnections have been developed. Such interconnections generally consist of multiple layers of metallic conductor lines embedded in a low dielectric constant material. The dielectric constant in such material has a very important influence on the performance of an integrated circuit. Materials having low dielectric constants (i.e., below 2.5) are desirable because they allow faster signal velocity and shorter cycle times. In general, low dielectric constant materials reduce capacitive effects in integrated circuits, which frequently leads to less cross talk between conductor lines, and allows for lower voltages to drive integrated circuits.
Low dielectric constant materials can be characterized as predominantly inorganic or organic. Inorganic oxides often have dielectric constants between 2.5 and 4, which tends to become problematic when device features in integrated circuits are smaller than 1 .mu.m. Organic polymers include epoxy networks, cyanate ester resins, polyarylene ethers, and polyimides. Epoxy networks frequently show disadvantageously high dielectric constants at about 3.8-4.5. Cyanate ester resins have relatively low dielectric constants between approximately 2.5-3.7, but tend to be rather brittle, thereby limiting their utility. Polyimides and polyarylene ethers, have shown many advantageous properties including high thermal stability, ease of processing, low stress, low dielectric constant and high resistance, and such polymers are therefore frequently used as alternative low dielectric constant polymers.
With respect to other properties, desirable dielectrics should also be free from moisture and out-gassing problems, have suitable adhesive and gap-filling qualities, and have suitable dimensional stability towards thermal cycling, etching, and CMP processes (i.e., chemical, mechanical, polishing). Preferred dielectrics should also have Tg values (glass transition temperatures) of at least 300.degree. C., and preferably 400.degree. C. or more.
The demand for materials having dielectric constant lower than 2.5 has led to the development of dielectric materials with designed-in nanoporosity. Since air has a dielectric constant of about 1.0, a major goal is to reduce the dielectric constant of nanoporous materials down towards a theoretical limit of 1. Several approaches are known in the art for fabricating nanoporous materials. In one approach, small hollow glass spheres are introduced into a material. Examples are given in U.S. Pat. No. 5,458,709 to Kamezaki and U.S. Pat. No. 5,593,526 to Yokouchi. However, the use of small, hollow glass spheres is typically limited to inorganic silicon-containing polymers.
In another approach, a thermostable polymer is blended with a thermolabile (thermally decomposable) polymer. The blended mixture is then crosslinked and the thermolabile portion thermolyzed. Examples are set forth in U.S. Pat. No. 5,776,990 to Hedrick et al. Alternatively, thermolabile blocks and thermostable blocks alternate in a single block copolymer, or thermostable blocks and thermostable blocks carrying thermolabile portions are mixed and polymerized to yield a copolymer. The copolymer is subsequently heated to thermolyze the thermolabile blocks. Dielectrics with k-values of 2.2, or less have been produced employing thermolabile portions. However, many difficulties are encountered utilizing mixtures of thermostable and thermolabile polymers. For example, in some cases distribution and pore size of the nanovoids is difficult to control. In addition, the temperature difference between thermal decomposition of the thermolabile group and the glass transition temperature (Tg) of the dielectric is relatively low. Still further, an increase in the concentration of thermolabile portions in a dielectric generally results in a decrease in mechanical stability.
In yet another approach, a polymer is formed from a first solution in the presence of microdroplets of a second solution, where the second solution is essentially immiscible with the first solution. During polymerization, rnicrodroplets are entrapped in the forming polymeric matrix. After polymerization, the microdroplets of the second solution are evaporated by heating the polymer to a temperature above the boiling point of the second solution, thereby leaving nanovoids in the polymer. However, generating nanovoids by evaporation of microdroplets suffers from several disadvantages. Evaporation of fluids from polymeric structures tends to be an incomplete process that may lead to undesired out-gassing, and potential retention of moisture. Furthermore, many solvents have a relatively high vapor pressure, and methods using such solvents therefore require additional heating or vacuum treatment to completely remove such solvents. Moreover, employing microdroplets to generate nanovoids often allows little control over pore size and pore distribution.
These problems are addressed in copending applications, Ser. Nos.: 60/128465; 60/128533; 60/128534; 60/128493; and 60/133218. In these applications, it is disclosed that nanoporous materials can be fabricated a) from polymers having backbones with reactive groups used in crosslinking; b) from polymer strands having backbones that are crosslinked using ring structures; and c) from stable, polymeric template strands having reactive groups that can be used for adding thermolabile groups or for crosslinking; d) by depositing cyclic oligomers on a substrate layer of the device, including the cyclic oligomers in a polymer, and crosslinking the polymer to form a crosslinked polymer; and e) by using a dissolvable phase to form a polymer.
Regardless of the approach used to introduce the voids, structural problems are frequently encountered in fabricating and processing nanoporous materials. Among other things, increasing the porosity beyond a critical extent (generally about 30% in the known nanoporous materials) tends to cause the porous materials to be weak and in some cases to collapse. Collapse can be prevented to some degree by adding crosslinking additives that couple thermostable portions with other thermostable portions, thereby producing a more rigid network. However, the porous material, even after cross-linking, can lose mechanical strength as the porosity increases, and the material will be unable to survive during integration of the dielectric film to a circuit.
Therefore, there is a need to provide methods and compositions to produce nanoporous low dielectric materials that combine the maximum allowable porosity with the maximum allowable durability and film strength.